Research On Option Pricing Of Continuous Railway Bulk Freight Based On Deep Learning And Transfer Learning

With the gradual development of railway freight reform,it is inevitable to study the freight rate of railway bulk goods as an important role in freight market.However,the original rigid freight rate system can not adapt to the current fierce market competition,scientific railway bulk freight rate system should reflect the value of transportation,supply and demand,and adapt to market competition.Therefore,under the premise of diversified demand of transportation market,it is a key problem for railway transportation enterprises to enrich and perfect the market-oriented pricing theory of railway bulk cargo transportation.Accordingly,this paper takes into account the inherent needs of the railway bulk transport market,explores the existing tariff system in depth,and develops and designs a continuous railway bulk cargo freight option pricing model based on deep learning and transfer learning to meet the fierce competition in the bulk cargo transport market.Firstly,on the premise of clarifying the definition of railway bulk cargo freight options,the trading process and analysing the influencing factors affecting the option price,a continuous railway bulk cargo freight options pricing model under deep learning,the CNN＿BI-LSTM parallel network model,is constructed,which combines a convolutional neural network(CNN),a bidirectional long and short-term memory neural network(BI-LSTM)and fully connected neural network(FCNN),and uses parallel learning mechanisms to achieve multi-feature data fusion in order to process complex option data.Data sets with sufficient option data are selected for example study,the results show that the proposed CNN＿BI-LSTM parallel network model has higher forecasting performance compared with the Black＿Scholes option pricing model,BP neural network,CNN,and long and short-term memory neural network(LSTM).Secondly,to address the situation that there is a scarcity of option data in the freight market and the CNN＿BI-LSTM parallel network model struggles to guarantee the prediction accuracy,a continuous railway bulk cargo freight option pricing model based on Two-stage TrAdaBoost.R2＿CNN＿BI-LSTM is proposed,which combines knowledge transfer of transfer learning(Two-stage TrAdaBoost.R2 algorithm)and regression capability of CNN＿BI-LSTM parallel network model to process limited options data.Data sets with scarce option data are selected for example study,the results show that the proposed Two-stage TrAdaBoost.R2＿CNN＿BI-LSTM model has better results compared with Support Vector Machine Regression(SVR),Random Forest(RF)and CNN＿BI-LSTM,and can effectively solve the pricing problem of option data scarcity.20 figures,15 tables and 84 references.